{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:6LDABG62QF4WJBT45EICYC5VC3","short_pith_number":"pith:6LDABG62","schema_version":"1.0","canonical_sha256":"f2c6009bda817964867ce9102c0bb516e774e0626183a0d3274ac5cafdcb398a","source":{"kind":"arxiv","id":"2406.17746","version":2},"attestation_state":"computed","paper":{"title":"Recite, Reconstruct, Recollect: Memorization in LMs as a Multifaceted Phenomenon","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Alvin Deng, Christopher A. Choquette-Choo, Jacob Ray Fuehne, Jaydeep Borkar, Jyothir S V, Katherine Lee, Kyle O'Brien, Mohammad Aflah Khan, Naomi Saphra, Stella Biderman, Tracy Ke, USVSN Sai Prashanth","submitted_at":"2024-06-25T17:32:16Z","abstract_excerpt":"Memorization in language models is typically treated as a homogenous phenomenon, neglecting the specifics of the memorized data. We instead model memorization as the effect of a set of complex factors that describe each sample and relate it to the model and corpus. To build intuition around these factors, we break memorization down into a taxonomy: recitation of highly duplicated sequences, reconstruction of inherently predictable sequences, and recollection of sequences that are neither. We demonstrate the usefulness of our taxonomy by using it to construct a predictive model for memorization"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2406.17746","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-25T17:32:16Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"21ef64110452ec5f0af2304444104549dff27b619993fb8f9bd1b753f4a9c18f","abstract_canon_sha256":"a1a915eeffd79de6d07bb74dd79caf5db3c4d3cae6b9a5efb668d4b3b300d765"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:59:59.084835Z","signature_b64":"tlTTMlwDiiMwIZu4seRLDz+vTazo0nIse40lyEuby/BywwhiWI4WAJ8TB/7IVfGgtfQbrZ9S8Dm1GaAfKzt2Aw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f2c6009bda817964867ce9102c0bb516e774e0626183a0d3274ac5cafdcb398a","last_reissued_at":"2026-07-05T10:59:59.084328Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:59:59.084328Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Recite, Reconstruct, Recollect: Memorization in LMs as a Multifaceted Phenomenon","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Alvin Deng, Christopher A. Choquette-Choo, Jacob Ray Fuehne, Jaydeep Borkar, Jyothir S V, Katherine Lee, Kyle O'Brien, Mohammad Aflah Khan, Naomi Saphra, Stella Biderman, Tracy Ke, USVSN Sai Prashanth","submitted_at":"2024-06-25T17:32:16Z","abstract_excerpt":"Memorization in language models is typically treated as a homogenous phenomenon, neglecting the specifics of the memorized data. We instead model memorization as the effect of a set of complex factors that describe each sample and relate it to the model and corpus. To build intuition around these factors, we break memorization down into a taxonomy: recitation of highly duplicated sequences, reconstruction of inherently predictable sequences, and recollection of sequences that are neither. We demonstrate the usefulness of our taxonomy by using it to construct a predictive model for memorization"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.17746","kind":"arxiv","version":2},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2406.17746/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2406.17746","created_at":"2026-07-05T10:59:59.084390+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.17746v2","created_at":"2026-07-05T10:59:59.084390+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.17746","created_at":"2026-07-05T10:59:59.084390+00:00"},{"alias_kind":"pith_short_12","alias_value":"6LDABG62QF4W","created_at":"2026-07-05T10:59:59.084390+00:00"},{"alias_kind":"pith_short_16","alias_value":"6LDABG62QF4WJBT4","created_at":"2026-07-05T10:59:59.084390+00:00"},{"alias_kind":"pith_short_8","alias_value":"6LDABG62","created_at":"2026-07-05T10:59:59.084390+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2505.12546","citing_title":"Extracting memorized pieces of (copyrighted) books from open-weight language models","ref_index":213,"is_internal_anchor":false},{"citing_arxiv_id":"2605.03547","citing_title":"Erase Persona, Forget Lore: Benchmarking Multimodal Copyright Unlearning in Large Vision Language Models","ref_index":33,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/6LDABG62QF4WJBT45EICYC5VC3","json":"https://pith.science/pith/6LDABG62QF4WJBT45EICYC5VC3.json","graph_json":"https://pith.science/api/pith-number/6LDABG62QF4WJBT45EICYC5VC3/graph.json","events_json":"https://pith.science/api/pith-number/6LDABG62QF4WJBT45EICYC5VC3/events.json","paper":"https://pith.science/paper/6LDABG62"},"agent_actions":{"view_html":"https://pith.science/pith/6LDABG62QF4WJBT45EICYC5VC3","download_json":"https://pith.science/pith/6LDABG62QF4WJBT45EICYC5VC3.json","view_paper":"https://pith.science/paper/6LDABG62","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.17746&json=true","fetch_graph":"https://pith.science/api/pith-number/6LDABG62QF4WJBT45EICYC5VC3/graph.json","fetch_events":"https://pith.science/api/pith-number/6LDABG62QF4WJBT45EICYC5VC3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6LDABG62QF4WJBT45EICYC5VC3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6LDABG62QF4WJBT45EICYC5VC3/action/storage_attestation","attest_author":"https://pith.science/pith/6LDABG62QF4WJBT45EICYC5VC3/action/author_attestation","sign_citation":"https://pith.science/pith/6LDABG62QF4WJBT45EICYC5VC3/action/citation_signature","submit_replication":"https://pith.science/pith/6LDABG62QF4WJBT45EICYC5VC3/action/replication_record"}},"created_at":"2026-07-05T10:59:59.084390+00:00","updated_at":"2026-07-05T10:59:59.084390+00:00"}